MJO Event Sensitivities to Tropical SSTs in Observations, Low-Order, and Global Climate Models

Global climate models (GCMs) historically struggle with simulating many of the observed features of the Madden-Julian oscillation (MJO). While newer models have shown some success at producing sufficient intraseasonal variance, these models are typically evaluated as composites and on climatological timescales rather than the temporal organization and eastward propagation of individual MJOs. It thus becomes worthwhile to investigate how well GCMs simulate higher-order event statistics such as the number and length of individual MJO events. In addition, sensitivities of the MJO to the shape and strength of the tropical warm pool remain largely unknown, considering that recent observational studies obtain conflicting results pertaining to MJO event length and propagation speed for different phases of the El Niño-Southern Oscillation (ENSO) and Indian Ocean dipole (IOD).

This study presents an overview of individual MJO event statistics in observations and numerical models. Specifically, we examine those models with good MJO simulations from the recent MJO Task Force and GEWEX Atmosphere System Study (GASS) intercomparison project, in addition to the low-order MJO “skeleton model” developed by Majda and Stechmann (2009, 2011). The stochastic version of the skeleton model produces realistic MJO event statistics for such a simple model, albeit with fewer individual events that persist slightly longer than observed. In contrast, most GCMs simulate too many MJOs given the challenges of eastward propagation and MJO maintenance in climate models.

Observed MJO events are shorter-lived during El Niño when subset by season and ENSO phase. There is also a relative increase in the number of individually occurring MJOs compared to neutral or La Niña conditions. The frequency of MJOs and corresponding event lengths are less sensitive to tropical sea surface temperatures (SSTs) in the low-order model, though the skeleton model demonstrates additional eastward propagation with enhanced MJO variance over the central Pacific when forced by El Niño SSTs. Similar results are found when forcing the skeleton model with ENSO-varying profiles of radiative cooling and low-level moistening, consistent with observations. Understanding the statistical influence of natural climate variations in the large-scale tropical forcing may help lead towards improved simulations of the MJO in global climate and weather forecast models.